Can generative AI outshine the human imagination? It’s a tricky (if scary) question, and studies so far have found mixed results. Lynn Wu, an associate professor at The Wharton School, and her colleagues decided to measure AI’s effect on product innovation by analyzing metrics like patents and revenue.
Related: This Entrepreneur Used AI to Transform Their Business and Create Multiple Revenue Streams — Here’s Exactly How They Did It
The research
Wu’s study was built on two kinds of research: a survey of 331 companies about their technology development practices, plus an analysis of more than 2,000 publicly traded firms’ productivity and new patents filed. Her goal was to identify two things: How are companies integrating AI? And what effect does using it have on innovating products with commercial value?
The results
AI was bad at creating “out of thin air” breakthroughs. “That’s a fundamental limitation,” Wu says. “When it does come up with something radically new, it could be an AI hallucination.” But AI is very good at another kind of innovation — combining existing ideas to create something new.
Many companies in the study used AI for this purpose, and were rewarded for it: They were 3% to 7% more productive with patents and revenue than firms that did not use these tools. AI was especially impactful at heavily siloed and decentralized companies. When different departments weren’t talking to each other, AI bridged the gap by making knowledge accessible to everyone on the team looking to develop a new product.
Related: Here’s Your Cheat Sheet of AI Tools That Actually Work, According to Real Entrepreneurs
What we’ve learned
Humans are good at linking three to five ideas in our heads. “But when you need to combine more elements, that’s when the machine can really help you,” Wu says. For example, if you want to dream up a new beverage that teens will go ape for, generative AI can mix and match thousands of preferences for that age group to find a concept that hits the mark.
How to use it
Want to innovate your own product? Tojin Eapen, a senior fellow at The Conference Board, developed a research-backed strategy based on linking: Ask generative AI to combine three random words (say, “phone,” “lava,” and “lobster”) into a new product concept. What you get back may not be so practical, but it can inspire your team to develop a more valuable idea. For a specific kind of product — say, footwear — you could start with your category (“shoe”) and add two other words that are random (“telescope”) or related (“safety”) and see what happens.
Before investing time or money in an idea, Eapen suggests getting AI to play devil’s advocate. You might ask: “What are 10 ways this idea could fail?” or “What will customers dislike about this product?” or “What are the three critical risks?”
As you use AI, however, always remember this: It might be helpful — but it’ll be helpful to your competitor in exactly the same ways. “So whether it’s proprietary data or something unique,” says Wu, “you better figure out your moat to protect what you’ve got.”
Related: 5 Key Ways to Create an Innovation Culture
